SB37 / SD37 / SE37 — out of space

An output dataset could not get more space. B37, D37 and E37 each fail for a different reason, so read the code before you change SPACE.

What happened

A step writing an output dataset ends with SB37, SD37 or SE37. An IEC message names the DD and the dataset.

Typical JES messages (illustrative)

IEC030I B37-04,IFG0554A,PAYJOB01,STEP030,OUTFILE,0A41,PRD123,
PROD.PAYROLL.DETAIL.D251008
IEF450I PAYJOB01 STEP030 - ABEND=SB37 U0000 REASON=00000004

D37 is reported with IEC031I and E37 with IEC032I.

What it means

A sequential or partitioned Dataset filled its current space and the system could not give it more. The three codes say why:

Code	What stopped the dataset growing	First thing to check
SB37	It needed more space on its current volume, could not get it, and could not continue on another volume.	Secondary size, volume count, and whether the output is far bigger than normal.
SD37	Primary space was used up and no secondary quantity was specified, so it could not extend at all.	The SPACE parameter: add a secondary quantity.
SE37	The maximum number of extents or volumes was reached. A PDS cannot span volumes, so this is typical for a full PDS.	For a PDS: compress or reallocate larger. For sequential: the volume count.

Typical causes

The data volume grew beyond what the allocation allowed.
No secondary space was coded (D37).
A small secondary quantity used up the extent limit — a basic-format sequential dataset can have at most 16 extents on a volume.
A PDS filled with dead space from replaced members (E37).
A program loop or duplicated input wrote far more records than usual.
The storage group or volumes the dataset can use are full.

Symptoms

The failing DD is always an output or an extend. Compare the record count with previous runs: a normal-size output points to SPACE; a huge one points to a loop or bad input. VSAM datasets report space problems through VSAM messages and status codes, not these abends. A full PDS directory is a different abend (SB14), not E37.

Where to look

JESMSGLG and JESYSMSG: the IEC03xI message names the DD, dataset and volume.
JESJCL: the SPACE, UNIT and DCB on the failing DD.
ISPF 3.4 with I: allocated, used and extents of the dataset.
Previous runs' output sizes in SDSF or the scheduler history.

How to diagnose

Read the exact code: B37, D37 or E37.
Find the DD and dataset from the IEC message.
Check the SPACE parameter and any data class that sets space.
Compare this run's record count with normal runs.
If the volume is the problem, ask storage administration about free space in the storage group.

How to fix

D37: add a secondary quantity.
B37 or E37 on sequential: raise primary to cover normal volume, and make the secondary large enough not to run out of extents.
E37 on a PDS: compress it (IEBCOPY or Z in 3.4) or reallocate it larger, or use a PDSE.
Add RLSE so unused space is returned at close.
Delete the partial output, or let the step's DISP handle it, then rerun the step.

Some SMS data classes allow extra volumes or space constraint relief; that is site configuration, so check with your storage team.

How to prevent

Size allocations from real data volumes with headroom, and review them as volumes grow.
Always code a sensible secondary quantity.
Prefer PDSE for libraries that change often.
Monitor large files with LISTCAT or 3.4 rather than waiting for the abend.

Production considerations

Interview question

What is the difference between SB37, SD37 and SE37?